1. Field of the Invention
The present invention relates generally to the screening of compound and combinatorial libraries. More particularly, the present invention relates to a method and apparatus for performing assays, particularly thermal shift assays.
2. Related Art
In recent years, pharmaceutical researchers have turned to combinatorial libraries as sources of new lead compounds for drug discovery. A combinatorial library is a collection of chemical compounds which have been generated, by either chemical synthesis or biological synthesis, by combining a number of chemical "building blocks" as reagents. For example, a combinatorial polypeptide library is formed by combining a set of amino acids in every possible way for a given compound length (i.e., the number of amino acids in a polypeptide compound). Millions of chemical compounds can theoretically be synthesized through such combinatorial mixing of chemical building blocks. Indeed, one investigator has observed that the systematic, combinatorial mixing of 100 interchangeable chemical building blocks results in the theoretical synthesis of 100 million tetrameric compounds or 10 billion pentameric compounds (Gordon, E. M. et al., J. Med. Chem. 37:1233-1251 (1994)).
The rate of combinatorial library synthesis is accelerated by automating compound synthesis and evaluation. For example, DirectedDiversity.RTM. is a computer based, iterative process for generating chemical entities with defined physical, chemical and/or bioactive properties. The DirectedDiversity.RTM. system is disclosed in U.S. Pat. No. 5,463,564, which is herein incorporated by reference in its entirety.
Once a library has been constructed, it must be screened to identify compounds which possess some kind of biological or pharmacological activity. To screen a library of compounds, each compound in the library is equilibrated with a target molecule of interest, such as an enzyme. A variety of approaches have been used to screen combinatorial libraries for lead compounds. For example, in an encoded library, each compound in a chemical combinatorial library can be made so that an oligonucleotide "tag" is linked to it. A careful record is kept of the nucleic acid tag sequence for each compound. A compound which exerts an effect on the target enzyme is selected by amplifying its nucleic acid tag using the polymerase chain reaction (PCR). From the sequence of the tag, one can identify the compound (Brenner, S. et al., Proc. Natl. Acad. Sci. USA 89:5381-5383 (1992)). This approach, however, is very time consuming because it requires multiple rounds of oligonucleotide tag amplification and subsequent electrophoresis of the amplification products.
A filamentous phage display peptide library can be screened for binding to a biotinylated antibody, receptor or other binding protein. The bound phage is used to infect bacterial cells and the displayed determinant (i.e., the peptide ligand) is then identified (Scott, J. K. et al., Science 249:386-390 (1990)). This approach suffers from several drawbacks. It is time consuming. Peptides which are toxic to the phage or to the bacterium cannot be studied. Moreover, the researcher is limited to investigating peptide compounds.
In International Patent Application WO 94/05394 (1994), Hudson, D. et al., disclose a method and apparatus for synthesizing and screening a combinatorial library of biopolymers on a solid-phase plate, in an array of 4.times.4 to 400.times.400. The library can be screened using a fluorescently labeled, radiolabeled, or enzyme-linked target molecule or receptor. The drawback to this approach is that the target molecule must be labeled before it can be used to screen the library.
A challenge presented by currently available combinatorial library screening technologies is that they provide no information about the relative binding affinities of different ligands for a receptor protein. This is true whether the process for generating a combinatorial library involves phage library display of peptides (Scott, J. K. et al., Science 249:386-390 (1990)), random synthetic peptide arrays (Lam, K. S. et al., Nature 354:82-84 (1991)), encoded chemical libraries (Brenner, S. et al., Proc. Natl. Acad. Sci. USA 89:5381-5383 (1992)), the method of Hudson (Intl. Appl. WO 94/05394), or most recently, combinatorial organic synthesis (Gordon, E. et al., J. Med. Chem. 37:1385-1399 (1994)).
To acquire quantitative binding data from the high throughput screening of ligand affinities for a target enzyme, researchers have relied on assays of enzyme activity. Enzymes lend themselves to high throughput screening because the effect of ligand binding can be monitored using kinetic assays. The experimental endpoint is usually a spectrophotometric change. Using a kinetic assay, most researchers use a two-step approach to lead compound discovery. First, a large library of compounds is screened against the target enzyme to determine if any of the library compounds are active. These assays are usually performed in a single concentration (between 10.sup.-4 -10.sup.-6 M) with one to three replicates. Second, promising compounds obtained from the first screen (i.e., compounds which display activity greater than a predetermined value) are usually re-tested to determine a 50% inhibitory concentration (IC.sub.50), an inhibitor association constant (K.sub.i), or a dissociation constant (K.sub.d). This two-step approach, however, is very labor intensive, time-consuming and prone to error. Each re-tested sample must either be retrieved from the original assay plate or weighed out and solubilized again. A concentration curve must then be created for each sample and a separate set of assay plates must be created for each assay.
There are other problems associated with the biochemical approach to high throughput screening of combinatorial libraries. Typically, a given assay is not applicable to more than one receptor. That is, when a new receptor becomes available for testing, a new assay must be developed. For many receptors, reliable assays are simply not available. Even if an assay does exist, it may not lend itself to automation. Further, if a K.sub.i is the endpoint to be measured in a kinetic assay, one must first guess at the concentration of inhibitor to use, perform the assay, and then perform additional assays using at least six different concentrations of inhibitor. If one guesses too low, an inhibitor will not exert its inhibitory effect at the suboptimal concentration tested.
In addition to the drawbacks to the kinetic screening approach described above, it is difficult to use the kinetic approach to identify and rank ligands that bind outside of the active site of the enzyme. Since ligands that bind outside of the active site do not prevent binding of spectrophotometric substrates, there is no spectrophotometric change to be monitored. An even more serious drawback to the kinetic screening approach is that non-enzyme receptors cannot be assayed at all.
Thermal protein unfolding, or thermal "shift," assays have been used to determine whether a given ligand binds to a target receptor protein. In a physical thermal shift assay, a change in a biophysical parameter of a protein is monitored as a function of increasing temperature. For example, in calorimetric studies, the physical parameter measured is the change in heat capacity as a protein undergoes temperature induced unfolding transitions. Differential scanning calorimetry has been used to measure the affinity of a panel of azobenzene ligands for streptavidin (Weber, P. et al., J. Am. Chem. Soc. 16:2717-2724 (1994)). Titration calorimetry has been used to determine the binding constant of a ligand for a target protein (Brandts, J. et al., American Laboratory 22:30-41 (1990)). The calorimetric approach, however, requires that the researcher have access to a calorimetric device. In addition, calorimetric technologies do not lend themselves to the high throughput screening of combinatorial libraries, three thermal scans per day are routine.
Like calorimetric technologies, spectral technologies have been used to monitor temperature induced protein unfolding (Bouvier, M. et al., Science 265:398-402 (1994); Chavan, A. J. et al., Biochemistry 33:7193-7202 (1994); Morton, A. et al., Biochemistry 1995:8564-8575 (1995)). The calorimetric and spectral thermal shift studies described above all share a common limitation. In each study, only one binding reaction was heated and assayed at a time. The single sample heating and assay configuration, as conventionally performed, has impeded the application of thermal shift technologies to high throughput screening of combinatorial libraries. Thus, there is a need for a thermal shift technology which can be used to screen combinatorial libraries, can be used to identify and rank lead compounds, and is applicable to all receptor proteins.
Thermal shift assays have been used to determine whether a ligand binds to DNA. Calorimetric, absorbance, circular dichroism, and fluorescence technologies have been used (Pilch, D. S. et al., Proc. Natl. Acad. Sci. U.S.A. 91:9332-9336 (1994); Lee, M. et al., J. Med. Chem. 36:863-870 (1993); Butour, J.-L. et al., Eur. J. Biochem. 202:975-980 (1991); Barcelo, F. et al., Chem. Biol. Interactions 74:315-324 (1990)). As used conventionally, however, these technologies have impeded the high throughput screening of nucleic acid receptors for lead compounds which bind with high affinity. Thus, there is a need for a thermal shift technology which can be used to identify and rank the affinities of lead compounds which bind to DNA sequences of interest.
When bacterial cells are used to overexpress exogenous proteins, the recombinant protein is often sequestered in bacterial cell inclusion bodies. For the recombinant protein to be useful, it must be purified from the inclusion bodies. During the purification process, the recombinant protein is denatured and must then be renatured. It is impossible to predict the renaturation conditions that will facilitate and optimize proper refolding of a given recombinant protein. Usually, a number of renaturing conditions must be tried before a satisfactory set of conditions is discovered. In a study by Tachibana et al, each of four disulfide bonds were singly removed, by site-directed mutagenesis, from hen lysozyme (Tachibana et al., Biochemistry 33:15008-15016 (1994)). The mutant genes were expressed in bacterial cells and the recombinant proteins were isolated from inclusion bodies. Each of the isolated proteins were renatured under different temperatures and glycerol concentrations. The efficacy of protein refolding was assessed in a bacteriolytic assay in which bacteriolytic activity was measured as a function of renaturing temperature. The thermal stability of each protein was studied using a physical thermal shift assay. In this study, however, only one sample reaction was heated and assayed at a time. The single sample heating and assay configuration prevents the application of thermal shift technologies to high throughput screening of a multiplicity of protein refolding conditions. Thus, there is a need for a thermal shift technology which can be used to rank the efficacies of various protein refolding conditions.
Over the past four decades, X-ray crystallography and the resulting atomic models of proteins and nucleic acids have contributed greatly to an understanding of structural, molecular, and chemical aspects of biological phenomena. However, crystallographic analysis remains difficult because there are not straightforward methodologies for obtaining X-ray quality protein crystals. Conventional methods cannot be used quickly to identify crystallization conditions that have highest probability of promoting crystallization (Garavito, R. M. et al., J. Bioenergtics and Biomembranes 28:13-27 (1996)). Even the use of factorial design experiments and successive automated grid searches (Cox, M. J., & Weber, P. C., J. Appl. Cryst. 20:366-373 (1987); Cox, M. J., & Weber, P. C., J. Crystal Growth 90:318-324 (1988)) do not facilitate rapid, high throughput screening of biochemical conditions that promote the crystallization of X-ray quality protein crystals. Moreover, different proteins are expected to require different conditions for protein crystallization, just as has been the experience for their folding (McPherson, A., In: Preparation and Analysis of Protein Crystals, Wiley Interscience, New York, (1982)). Conventional methods of determining crystallization conditions are cumbersome, slow, and labor intensive. Thus, there is a need for a rapid, high throughput technology which can be used to rank the efficacies of protein crystallization conditions.
Rapid, high throughput screening of combinatorial molecules or biochemical conditions that stabilize target proteins in thermal shift assays would be facilitated by the simultaneous heating of many samples. To date, however, thermal shift assays have not been performed that way. Instead, the conventional approach to performing thermal shift assays has been to heat and assay only one sample at a time. That is, researchers conventionally 1) heat a sample to a desired temperature in a heating apparatus; 2) assay a physical change, such as absorption of light or change in secondary, tertiary, or quaternary protein structure; 3) heat the samples to the next highest desired temperature; 4) assay for a physical change; and 5) continue this process repeatedly until the sample has been assayed at the highest desired temperature.
This conventional approach is disadvantageous for at least two reasons. First, this approach is labor intensive. Second, this approach limits the speed with which thermal shift screening assays can be performed and thereby precludes rapid, high-throughput screening of combinatorial molecules binding to a target receptor and biochemical conditions that stabilize target proteins. Thus, there is a need for an apparatus capable of performing rapid, high-throughput thermal shift assays that will be suitable for all receptors, including reversibly folding proteins.